Seeking the Supergenius
Seeking the Supergenius: The Life of a High-Frequency Trader
Q&A with Igor Tulchinsky, founder & CEO, WorldQuant
March 22, 2010
What types of securities lend themselves to high-frequency trading?
"They all lend themselves," says Igor Tulchinsky, 43, founder and CEO of WorldQuant, LLC., one of the larger firms practicing high-frequency algorithmic trading.
Headquartered in Greenwich, Conn., the company employs about 100 people and has a New York office, affiliate offices in Beijing and Bangalore and plans to open offices in Tel Aviv and London.
Tulchinsky spoke with technology reporter John Dodge about what life is like in a profession in which high-speed decision-making and trading operations are still being assessed by lawmakers and regulators (see sidebar). And where the approach to trading is not always that well understood in the securities industry itself.
SIN: Describe WorldQuant and how it operates.
Tulchinsky: We're a spin-off from Millennium Partners in 2007, and I guess if one word defines us, we're very quantitative- and process-driven, emphasizing efficiency and automated decision-making. We have a big research staff here in the U.S. and internationally. We're not a classic high-frequency trading firm. We're mixed, [doing] a little bit of high-frequency trading and a lot of statistical arbitrage. We trade mostly equities and futures worldwide. We have all kinds of things beneath the hood here.
What is statistical arbitrage (stat-arb)?
We develop mathematical models which describe how various instruments move and how to profit from these movements. Some are developed by researchers, some by traders and some are developed by machines scouring historical data day and night looking for patterns and inefficiencies. What we do is always simulated and back-tested. There's really no human intervention going on here. We strive to quantify and automate everything, so as a result, it's very quiet. You could fall asleep here.
The money-making part does not really come from research in the traditional sense. Moneymaking comes from statistical research, in our case.
In the case of stat-arb firms that are pure high-frequency firms, however, it comes from being connected to the exchanges and having very fast access, so you can act on the same alphas that everybody else has, but you're a little bit faster.
Who are your investors?
Right now, I am under contract and am not able to talk about who the investors are.>
Does that mean you are managing someone else's money? Is that the way most high-frequency trading firms operate?
I don't know if it's the way most operate. There's a lot of little small independent shops (see chart.) You do not need much capital [to start] high-frequency trading. It's very easy to set up a shop by yourself. We manage a fairly substantial amount of money, so it's a different story.
How reliable are these models?
They are reliable for a while and work for a few months and sometimes for a year and then stop working for reasons we don't really know. That's why a strong research function into new models is crucial, to keep finding new things.
Describe these models.